A survey of digital video passive forensics techniques

The video forgery passive forensics can directly distinguish the authenticity and integrity of the obtained video by utilizing the tampering traces without the aid of any prior axillary information such as digital watermark or perceptual hash signature. Thus, it has better adaptability and has become an important topic in the field of information security. In this survey, the characteristics of universal video forgery operations, their left tampered traces and influences to video passive forensics were in-depth analyzed. From two aspects of forensics strategy and techniques used, we discuss typical algorithms and methods in passive video forensics, including video source identification based, forgeries trails based, deep learning based, original video feature representation based techniques. Each of these categories of video forensics is summarized in detail, along with a critical analysis of the state of the art. The potential research directions and challenges of video passive forensics field are also investigated in detail.